Re: [PATCH v2 2/3] Y2038: nptl: Provide futex_abstimed_wait64 supporting 64 bit time

[Adhemerval Zanella <adhemerval.zanella@linaro.org>]

On 30/09/2020 09:47, Lukasz Majewski wrote:
> Hi Adhemerval,
> 
>> On 19/09/2020 10:07, Lukasz Majewski wrote:
>>> This is the helper function, which uses struct __timespec64
>>> to provide 64 bit absolute time to futex syscalls.
>>>
>>> The aim of this function is to move convoluted pre-processor
>>> macro code from sysdeps/nptl/lowlevellock-futex.h to C
>>> function in futex-internal.c
>>>
>>> The futex_abstimed_wait64 function has been put into a separate
>>> file on the purpose - to avoid issues apparent on the m68k
>>> architecture related to small number of available registers (there
>>> is not enough registers to put all necessary arguments in them if
>>> the above function would be added to futex-internal.h with
>>> __always_inline attribute).
>>>
>>> Additional precautions for m68k port have been taken - the
>>> futex-internal.c file will be compiled with -fno-inline flag.  
>>
>> LGTM with a nit regarding style and a clarification about 
>> '-fno-inline'.
> 
> Please find the explanation below.
> 
>>
>>>
>>> ---
>>> Changes for v2:
>>> - Handle the case when *abstime pointer is NULL
>>> ---
>>>  sysdeps/nptl/futex-internal.c         | 70
>>> +++++++++++++++++++++++++++ sysdeps/nptl/futex-internal.h         |
>>>  6 +++ sysdeps/unix/sysv/linux/m68k/Makefile |  2 +
>>>  3 files changed, 78 insertions(+)
>>>
>>> diff --git a/sysdeps/nptl/futex-internal.c
>>> b/sysdeps/nptl/futex-internal.c index 3366aac162..3211b4c94f 100644
>>> --- a/sysdeps/nptl/futex-internal.c
>>> +++ b/sysdeps/nptl/futex-internal.c
>>> @@ -45,6 +45,29 @@ __futex_abstimed_wait_cancelable32 (unsigned
>>> int* futex_word, abstime != NULL ? &ts32 : NULL,
>>>                                    NULL /* Unused.  */,
>>> FUTEX_BITSET_MATCH_ANY); }
>>> +
>>> +static int
>>> +__futex_abstimed_wait32 (unsigned int* futex_word,
>>> +                         unsigned int expected, clockid_t clockid,
>>> +                         const struct __timespec64* abstime,
>>> +                         int private)
>>> +{
>>> +  struct timespec ts32;
>>> +
>>> +  if (abstime != NULL && ! in_time_t_range (abstime->tv_sec))
>>> +    return -EOVERFLOW;
>>> +
>>> +  unsigned int clockbit = (clockid == CLOCK_REALTIME) ?
>>> +    FUTEX_CLOCK_REALTIME : 0;
>>> +  int op = __lll_private_flag (FUTEX_WAIT_BITSET | clockbit,
>>> private); +
>>> +  if (abstime != NULL)
>>> +    ts32 = valid_timespec64_to_timespec (*abstime);
>>> +
>>> +  return INTERNAL_SYSCALL_CALL (futex, futex_word, op, expected,
>>> +                                abstime != NULL ? &ts32 : NULL,
>>> +                                NULL /* Unused.  */,
>>> FUTEX_BITSET_MATCH_ANY); +}
>>>  #endif
>>>  
>>>  int  
>>
>> Ok.
>>
>>> @@ -97,3 +120,50 @@ __futex_abstimed_wait_cancelable64 (unsigned
>>> int* futex_word, futex_fatal_error ();
>>>      }
>>>  }
>>> +
>>> +int
>>> +__futex_abstimed_wait64 (unsigned int* futex_word, unsigned int
>>> expected,
>>> +                         clockid_t clockid,
>>> +                         const struct __timespec64* abstime, int
>>> private) +{
>>> +  unsigned int clockbit;
>>> +  int err;
>>> +
>>> +  /* Work around the fact that the kernel rejects negative timeout
>>> values
>>> +     despite them being valid.  */
>>> +  if (__glibc_unlikely ((abstime != NULL) && (abstime->tv_sec <
>>> 0)))
>>> +    return ETIMEDOUT;
>>> +
>>> +  if (! lll_futex_supported_clockid(clockid))
>>> +    return EINVAL;  
>>
>> Space before parentheses. 
> 
> Ok. Fixed.
> 
>>
>>> +
>>> +  clockbit = (clockid == CLOCK_REALTIME) ? FUTEX_CLOCK_REALTIME :
>>> 0;
>>> +  int op = __lll_private_flag (FUTEX_WAIT_BITSET | clockbit,
>>> private); +
>>> +  err = INTERNAL_SYSCALL_CALL (futex_time64, futex_word, op,
>>> expected,
>>> +                               abstime, NULL /* Unused.  */,
>>> +                               FUTEX_BITSET_MATCH_ANY);
>>> +#ifndef __ASSUME_TIME64_SYSCALLS
>>> +  if (err == -ENOSYS)
>>> +    err = __futex_abstimed_wait32 (futex_word, expected,
>>> +                                   clockid, abstime, private);
>>> +#endif
>>> +  switch (err)
>>> +    {
>>> +    case 0:
>>> +    case -EAGAIN:
>>> +    case -EINTR:
>>> +    case -ETIMEDOUT:
>>> +      return -err;
>>> +
>>> +    case -EFAULT: /* Must have been caused by a glibc or
>>> application bug.  */
>>> +    case -EINVAL: /* Either due to wrong alignment, unsupported
>>> +		     clockid or due to the timeout not being
>>> +		     normalized. Must have been caused by a glibc
>>> or
>>> +		     application bug.  */
>>> +    case -ENOSYS: /* Must have been caused by a glibc bug.  */
>>> +    /* No other errors are documented at this time.  */
>>> +    default:
>>> +      futex_fatal_error ();
>>> +    }
>>> +}  
>>
>> Ok.
>>
>>> diff --git a/sysdeps/nptl/futex-internal.h
>>> b/sysdeps/nptl/futex-internal.h index 7f3910ad98..1ba0d61938 100644
>>> --- a/sysdeps/nptl/futex-internal.h
>>> +++ b/sysdeps/nptl/futex-internal.h
>>> @@ -529,4 +529,10 @@ __futex_abstimed_wait_cancelable64 (unsigned
>>> int* futex_word, const struct __timespec64* abstime,
>>>                                      int private) attribute_hidden;
>>>  
>>> +int
>>> +__futex_abstimed_wait64 (unsigned int* futex_word, unsigned int
>>> expected,
>>> +                         clockid_t clockid,
>>> +                         const struct __timespec64* abstime,
>>> +                         int private) attribute_hidden;
>>> +
>>>  #endif  /* futex-internal.h */  
>>
>> Ok.
>>
>>> diff --git a/sysdeps/unix/sysv/linux/m68k/Makefile
>>> b/sysdeps/unix/sysv/linux/m68k/Makefile index
>>> be40fae68a..65164c5752 100644 ---
>>> a/sysdeps/unix/sysv/linux/m68k/Makefile +++
>>> b/sysdeps/unix/sysv/linux/m68k/Makefile @@ -21,3 +21,5 @@
>>> sysdep-dl-routines += dl-static sysdep-others += lddlibc4
>>>  install-bin += lddlibc4
>>>  endif
>>> +
>>> +CFLAGS-futex-internal.c += -fno-inline
>>>   
>>
>> Do we still need it?
> 
> Unfortunately, yes. The m68k architecture has the issue with properly
> compiling this code (in futex-internal.c) as it has very limited number
> of registers (8 x 32 bit IIRC).
> 
> I did some investigation and it looks like static inline
> in_time_t_range() function affects the compiler capabilities to
> generate correct code.
> 
> It can be easily inlined on any architecture but m68k.
> 
> As m68k has issues with this code compilation - it is IMHO better to
> disable inlining (-fno-inline) on this particular port.
> 
> As a result we would have slower execution only on relevant functions,
> but 64 bit time support would be provided. 

I recall this was need on first patch for the cancellable 64-bit
futex_abstimed_wait where it embedded the 32-bit fallback in the 
__futex_abstimed_wait_cancelable64. And my suggestion to move it to 
an external function seems to avoid the extra compiler flag.

This patchset uses the same strategy and I checked both patches
and it seems that -fno-inline is not required to build m68k.